cinemart-shinjuku

name	cinemart_shinjuku
description	Platform rules, page structure, and date extraction for the シネマート新宿 (Cinemart Shinjuku) scraper
applyTo	scraper/sources/cinemart_shinjuku.py

Cinemart Shinjuku Scraper — Platform Profile

Site Overview

Field	Value
Site URL	https://www.cinemart.co.jp/theater/shinjuku/
Tagline	アジアをもっと好きになる (Asia-focused)
Type	Static HTML — no JS rendering required
Auth	None
Rate Limit	time.sleep(0.5) between detail requests is sufficient
Source Name	cinemart_shinjuku
Source ID Format	cinemart_shinjuku_{6-digit-number} (e.g. cinemart_shinjuku_002491)
--source key	cinemart_shinjuku

Cinemart Shinjuku (新宿文化ビル6F・7F) is an Asia-focused art cinema. It screens Korean, Chinese, and other Asian films in addition to Taiwan films — a Taiwan keyword filter is always required.

Field Mappings

Event Field	Source
raw_title	First <h2> in <main>
start_date	First <p> in <main>: M月D日(曜)ロードショー
end_date	None for regular releases; = start_date for 1日限定上映
location_name	Fixed: シネマート新宿
location_address	Fixed: 東京都新宿区新宿3丁目13番3号 新宿文化ビル6F・7F
official_url	First external <a> with link text matching オフィシャルサイト / 公式サイト / official site in <main>
is_paid	Always True
price_info	None (no structured price on detail page)
category	["movie"]
source_url	Movie detail page URL

Page Structure

Listing page (/theater/shinjuku/movie/)

<a href="002491.html">  ← relative 6-digit link
  <h2>霧のごとく</h2>
  ... description text ...
</a>

• Selector: a[href] where href matches ^\d{6}\.html$
• Resolve to full URL via urljoin(_LISTING_URL, href)
• Both nowshowing and past/upcoming movies appear on this single page

Detail page (/theater/shinjuku/movie/NNNNNN.html)

<main>
  <p>5月8日(金)ロードショー</p>       ← FIRST <p> = release date
  <h2>霧のごとく</h2>                  ← Movie title
  <p>description...</p>
  <p>あらすじ content...</p>
  ...
  <h3>監督</h3>
  <h3>キャスト</h3>
  ...
  <dl>営業時間...</dl>                  ← SIDEBAR STARTS HERE — stop collecting
  <dl>劇場案内 新宿区新宿3丁目...</dl>

• Stop collecting <p> content when text contains _VENUE_STOP_WORDS
• Sidebar stop words: "オープン時間", "先売り指定席", "次週タイムスケジュール", "新宿区新宿3丁目", "東京メトロ", "詳しくはこちら"

Taiwan Relevance Filter

Keywords: ["台湾", "Taiwan", "臺灣", "金馬"]

• Fast pre-filter: Check listing link text (cheap, no HTTP request)
• Full filter: Check entire <main> text on detail page

金馬 (Golden Horse Award) is included as a Taiwan-specific indicator.

Date Format and Parsing

First <p> in <main> contains the release date:

Format	Parsed
5月8日(金)ロードショー	start=5/8, end=None (ongoing)
5月28日（木）1日限定上映	start=end=5/28 (single day)
4月17日(金)より限定開催！	start=4/17, end=None

Year inference: month < today.month - 3 → use today.year + 1.

Source ID

Extract the 6-digit number from the URL:

• https://...shinjuku/movie/002491.html → cinemart_shinjuku_002491
• Numbers are stable — used as dedup key across runs

Troubleshooting

Problem	Cause	Fix
raw_description includes sidebar info	_VENUE_STOP_WORDS not matching	Check actual sidebar <p> text; update stop words
No events despite Taiwan movies present	Taiwan keywords not in listing text	Check listing <a> text; add keyword to _TAIWAN_KEYWORDS
start_date is None	First <p> format changed	Check first <p> text on detail page; update regex in _parse_release_date
Listing returns 0 links	URL changed or ^\d{6}\.html$ pattern wrong	Check actual <a href> values on listing page

Phase 2 — Weekly Schedule Enrichment (added 2026-05-02)

After Phase 1 collects events via the listing/detail pages, Phase 2 enriches start_date, end_date, and business_hours from the online ticket schedule.

Field	Value
Schedule URL	https://cinemart.cineticket.jp/theater/shinjuku/schedule
Function	_parse_schedule_page(session) → dict[normalized_title, ScheduleEntry]
Fuzzy match	_normalize_title(title) strips 【4K上映】, 〈特別版〉 etc. before matching
business_hours format	"5/1（金）14:00・18:30\n5/2（土）10:30\n..." (one line per day)
Failure safe	try/except wraps Phase 2 entirely — Phase 1 results unaffected if schedule fails

_normalize_title rules

• Remove 【...】 and 〈...〉 bracketed modifiers (e.g. 【4K上映】, 〈特別版〉)
• Remove trailing whitespace
• Comparison is case-sensitive (Japanese titles don't have case variation)

Phase 2 matching algorithm

1. Build a dict[normalized_title, ScheduleEntry] from the schedule page
2. For each Phase 1 event, normalize raw_title and look up in the dict
3. If found: overwrite start_date, end_date, business_hours
4. If not found: keep Phase 1 values (some films may not appear in online ticketing)

lookup_movie_titles Integration

• Import: from movie_title_lookup import lookup_movie_titles (not relative)
• Call after title is confirmed: name_zh, name_en = lookup_movie_titles(title)
• Pass values to Event: name_zh=name_zh, name_en=name_en
• Silent failure — returns (None, None) if eiga.com lookup fails

Pending Rules

Add new lessons here after debugging sessions.